Looks like Intel is finally turning up the heat.

[Keysplayr]

Originally posted by: pm

And every month/year/decade somebody finds a way to do something faster and better than before. So who's to say that the concept to fab process isn't considerably shorter than it was say 10 years ago. Anyway, Intel should have "reacted" a lot sooner than they did. They finally are, but it took them sooooo long.

Actually, if anything it's getting longer from concept to shipment for microprocessors. Fabrication takes longer (more steps), mask prep takes longer (fracture, OPC, etc.), design teams are larger and more cumbersome, the fab process is introducing new issues and backside debug (C3) is a lot harder and slower than front-side debug (wirebond). 10 years ago the design cycle was shorter than it is now.

So your point is that Intel's management didn't react quickly enough - and my point is that it takes 4-5 years to create a new design. How fast can you react when it takes an absolute minimum of 3 years to bring a product to market?

So what you're saying is, Intel has done the best it could at all times? I don't think that at all if that is what you mean. It's not like Intel had to start from scratch to bring back the PIII, PM, Dothan, Yonah architecture now correct? How long ago did the Pentium 4 debut? 3rd or 4th Quarter of 2000. How long have Pentium M's been available in laptops? February 2003. When the Pentium 4 debuted, it flopped in every way possible except for Quake. WARNING!! Athlon blew its doors off and the Pentium III seriously outperformed it clock for clock. CLUE!!! Intel was 100% counting on the P4 to scale to 10GHz over time. That's great and all, but a big company like Intel should not have all its eggs in one basket, right? So they kept chugging along, raising clocks, adding cache, increasing FSB at an alarmingly slow rate. Not until 2.4GHz did Northwood scale well enough to pull ahead of AMD. VICTORY!!! (Not quite) Say hello to A64 with on die memory controller. How long has AMD had an on die memory controller now? September 2003. Intel had the resources to make sudden changes in their roadmaps, but the embarrassment factor was probably too high and they stuck it out, if not for any other reason, to keep stockholders faith in check. Didn't want to set off any alarm bells by saying, "Oops, we really did screw this pony royally.".

So, it seems to me you are saying Intel is going the fastest it possibly can. You are probably the only one who believes this. If they were, they would be looming large still over the Green Team. Intel got a dose of bad management. Correct me if I'm wrong. Please.

 

[dmens]

Originally posted by: Duvie
We already established that means nothing, genius...It was still a collosal failure. Chip has no headroom...stock HSF could barely keep it cool...I bet you Dell systems around the world are helping to heat ppls homes, cost them too much in electricity and probably throttling....

What's your point? Everyone knows 8xx was rushed (backpats all around).

Last time I checked, the success of a product was defined by how much money it makes.

 

[Duvie]

Originally posted by: dmens

Originally posted by: Duvie
We already established that means nothing, genius...It was still a collosal failure. Chip has no headroom...stock HSF could barely keep it cool...I bet you Dell systems around the world are helping to heat ppls homes, cost them too much in electricity and probably throttling....

What's your point? Everyone knows 8xx was rushed (backpats all around).

Last time I checked, the success of a product was defined by how much money it makes.

Lame argument for a lame product...I think we can leave it at that....

Plus that argument Intel fools can use all the time...When you have 70-80% of the market you are going to sell them.

No PATS on the back here, I was just indirectly commenting that PM is likely very right on the time it takes and noticeably should take to bring a successful (in terms of performance not sales alone) cpu to the market.

 

[Viditor]

Originally posted by: pm

Just to be clear, when I say Intel was "sleeping", that means not aggressive in the development department. They took seriously wrong turns toward long pipelines and Netburst, and by the time they realized it, KEPT GOING WITH IT. ????? Nobody was "awake" enough or in a proper decision making position to say, "Hey, we need to change the way we are doing things.

It's been my observation that few who aren't actually in the microprocessor industry have a really good understanding of just how long it takes to design, debug and ship a CPU.

It's about 4-5 years from start of a new microprocessor microarchitectural design to shipping a product.

QFT
While I'm not in the industry directly, I have been saying this for years...
Jerry Sanders actually put it best IMO, he compared it to a wierd kind of Russian Roulette:
"You put a gun to your head and pull the trigger," he says, "and four years later you find out if you blew your brains out."

Intel's problems today can be traced back to Barrett's decisions as CEO. Will Otellini fare better? It will be a year or tow before we know, but the hasty transition to NGMA was (IMHO) a smart one. They accelerated the transition when Tejas was cancelled in early 2004, and Otellini was voted in as CEO in November of 2004. I would bet that he had a lot of input on the transition. It's going to be another year or two before we see what kind of CEO Otellini really is, but so far it seems a good step forward (though I don't know how good an idea VIIV really is...).
There are still some serious problems for Intel to overcome...

1. In Aug, Intel cut back on the 915GL/PL chipset production. By November, there were big shortages in the low-end chipsets (Intel's biggest market at the time) and this allowed AMD to gobble up more marketshare than they should have. These kind of decisions have to stop...

2. The "power-bug" on Centrino laptops should never have gotten as far as it did. Sure it will be fixed, but it should have been caught much sooner.

3. Being forced to change platforms again for Conroe because Intel didn't revamp their VRM specs sooner will also be a costly mistake.

4. To overcome a rather poor Q4 earnings, Intel has given out several forward looking PRs on the NGMA. The effect of this has been that the share price hasn't fallen as much as it might have (good), and that many potential AMD customers have been "Osbourned" into waiting for the NGMA release in 9 months (also good). The other side though is very dangerous...if the NGMA isn't significantly better than AMD when it's released, there will be a very large number of dissapointed customers, and Intel's credibility will be significantly reduced (again). It's one Hell of a dice roll considering that Intel still doesn't really know the performance of what AMD are releasing!

Jerry Sanders once risked the entire company on Fab 30, and he got lucky. Intel's risk is nowhere near that big, but it IS quite substantial...we'll know at the end of this year if Intel will be able to maintain it's marketshare dominance, or if AMD will be able to pass their 30% marketshare goal and make it a true 2 horse race...

 

[Viditor]

Originally posted by: keysplayr2003
Intel got a dose of bad management. Correct me if I'm wrong. Please.

I agree completely...the big question is, have they got better management now??

 

[Marmion]

3. Being forced to change platforms again for Conroe because Intel didn't revamp their VRM specs sooner will also be a costly mistake.

I don't see how this could adversely affect Intel. I'd go out on a limb and say that most of Intel's CPU sales are to OEMs (Dell, HP etc), which would also mean that these OEMs would buy Intel chipsets with the CPUs. I don't seriously think that Dell/HP etc would ever considor using chipsets from today for their systems of 6 months away.
The enthusiest/upgrade market would be fairly small compared to OEM.

I also don't know why people would buy a motherboard now, expecting it to be compatable with NGMA. Do you expect your AMD K8 motherboards to work with AMD K10? Oh, and its going to happen again in 2007/8 when Intel and AMD change to DDR3.

Quite simply, it won't be a costly mistake. If the NGMA is better than AMD AM2, and the 65nm shrink, which it *should* be, then OEMs are going to be selling more and more of these CPUs, and enthusiests won't hold back if the performance is worth it.

All we can do is wait and see though...

 

[Viditor]

Originally posted by: Marmion

3. Being forced to change platforms again for Conroe because Intel didn't revamp their VRM specs sooner will also be a costly mistake.

I don't see how this could adversely affect Intel. I'd go out on a limb and say that most of Intel's CPU sales are to OEMs (Dell, HP etc), which would also mean that these OEMs would buy Intel chipsets with the CPUs. I don't seriously think that Dell/HP etc would ever considor using chipsets from today for their systems of 6 months away.
The enthusiest/upgrade market would be fairly small compared to OEM.

I also don't know why people would buy a motherboard now, expecting it to be compatable with NGMA. Do you expect your AMD K8 motherboards to work with AMD K10? Oh, and its going to happen again in 2007/8 when Intel and AMD change to DDR3.

Quite simply, it won't be a costly mistake. If the NGMA is better than AMD AM2, and the 65nm shrink, which it *should* be, then OEMs are going to be selling more and more of these CPUs, and enthusiests won't hold back if the performance is worth it.

All we can do is wait and see though...

You're right in that OEMs are Intel's main source of income...
However, OEMs have to design their systems on 3rd party motherboards the same as anyone else (they don't make their own...even Intel mobos are made by Foxconn). It is much more expensive (for the OEM) to require different designs and motherboards for different chips, and it makes inventory control more expensive/harder.

Well, my old 939 motherboard accepted my X2 with no difficulty...but the real problem is that up till now, Intel has been saying that they are compatable. This means that many future contracts will have to be changed, and people who are thinking of upgrading to Conroe shouldn't buy Intel yet. The easy compatability of the dual core was a huge win for AMD and helped them significantly with the Opteron rollout.

 

[Marmion]

I don't know this or not, but would there be any Intel press releases saying that current chipsets are compatable with Conroe?
I understand Merom will be as it uses the same chipset in the roadmaps as Yonah, but Conroe uses a faster bus, lower voltage and is completely different to the P4 architecture.
I think it is just rumours that the current 975x would be compatable, however I believe the 301 stepping is endeed compatable (obviously not with Conroe EE though).
At no point have I actually seen any official documents or press releases quoted saying that current desktop chipsets are compatable with NGMA, and I'm sure that motherboard makers and OEMs would be well aware of what chipsets will work with the NGMA (namely 946, 965 and 985 plus the 301 stepping of the 975 chipsets).

I still stand by the p;oint that OEMs won't be using current chipsets any way. They will most likely use the 965 or 946 for the mid-low end, and then 985 for when the EE comes out.
The new chipsets also have a new southbridge, so that is a semi-incentive to upgrade for enthusiasts.

I look forward to conroe anyway, might be upgrading my old Athlon XP 1800+ or Duron 500

 

[Viditor]

Sorry, I thought you'd seen the news blurb on this...

It's not the chipsets that are incompatable, it's Intel's specs for the motherboards...more specifically the VRM (Voltage Regulator Module).

 

[Cooler]

The real reason Intel was in the lead in market share is that AMD PR really sucks at best.

 

[Marmion]

So now its not really Intel's problem if you think about it this way:

Intel sets voltage spec for current CPUs.
-a) Motherboard maker either sticks to recommended spec or
-b) Motherboard maker expands vrm in anticipation of conroe

Result of a): More sales in future as enthusiasts see Conroe performance* and upgrade (new cpu/mobo)
Result of b): Less sales in future as enthusiasts see Conroe performance* and upgrade (new cpu, no need for mobo as already is compatable)

Bottom line: Option a) results in more sales and therefore greater profits for mobo maker = big check for CEO at end of year

*Note: I am not in anyway saying that Conroe is going to kick A64 X2s. I am not a fanboy If X2 still represents better value then considor this post to be void (And I'll buy an X2 instead)

 

[RichUK]

Intel I believe has to keep changing chipsets because the processors are somewhat reliant on the Northbridge, for supported features and functionality as the hub of all communication. So if something big changed on die for a given Intel processor, the latest Northbridge just wouldn?t have support for it (not in all circumstances).

This is another point where AMD excel?s, because AMD is not reliant on a Northbridge as this is all incorporated on die thus keeps changes on die, this then renders the Northbridge subsystem model redundant. If any major changes were to occur on die with AMD that is fine, because this affects the processor only and not the rest of the system. The latest change REV-F only requires a motherboard change because of the change of memory support, aka you can?t fit a DDR2 DIMM into a normal DDR slot, of course this change brings memory I/O change.

A thumbs up for AMD as this meant their whole s939 range could be compliant with pretty much all motherboards released (except a few via chipsets I think).

With s939 we saw single core through to dual core, and how many chipsets has Intel used in that same time frame!?!

Also reported in some articles socket F will see single > dual > quad core support on the same socket. That is quite an amazing concept for this to happen all on one socket. But I ask the question will AMD, in future support of socket F support DDR3 memory. I doubt AM2 will, but socket F has 1207 pins for future DDR3 expansion maybe?

 

[Viditor]

Rich has nailed it quite well...
AMD does have an advantage in cost to OEMs because of their design. The fewer mobo designs you need, the more you will save on total system cost.

My earlier point about it costing more is not that it won't work, it's that it should have been announced sooner to reduce the cost of system design by incorporating the wider VRM into current platforms. AMD was able to do this a full year ahead of the dual core release, and IMHO Intel was certainly capable of doing so last year as well. While I'm sure that mobo makers will indeed expand all vrm to be inclusive of Conroe from here on out, the cost of design and manufacturing line changes will be passed down to total system costs...

 

[Marmion]

I honestly don't know why its such a bid deal for OEMs. I mean, they are most certainly going to use the 965 or 946 chipsets with the conroes aren't they? They aren't going to spec their systems with old 975 chipsets which will have an outdated southbridge and are designed for enthusiasts and the Pressler EE anyway.
OK, for enthusiasts it is a pain to upgrade your mobo every time you upgrade your CPU. The answer is to either a) not upgrade so much, or b) go with the company that doesn't change their chipsets for every different CPU (ie AMD).

In the end, Intel is not going to lose out because they require everyone to use a new mobo. To be fair, it is a new architecture. They also have a new southbridge on the new chipsets aswell.

Intel will only lose out if Conroe is no better than the X2s. Somehow I think Conroe will be better than X2s, otherwise I see Intel going down fast.

All I'm trying to say is that a new chipset is not a big problem for Intel - that won't be their undoing in regards to point 3.
Everything else I agree with.

 

[Viditor]

Originally posted by: Marmion
I honestly don't know why its such a bid deal for OEMs. I mean, they are most certainly going to use the 965 or 946 chipsets with the conroes aren't they? They aren't going to spec their systems with old 975 chipsets which will have an outdated southbridge and are designed for enthusiasts and the Pressler EE anyway.
OK, for enthusiasts it is a pain to upgrade your mobo every time you upgrade your CPU. The answer is to either a) not upgrade so much, or b) go with the company that doesn't change their chipsets for every different CPU (ie AMD).

In the end, Intel is not going to lose out because they require everyone to use a new mobo. To be fair, it is a new architecture. They also have a new southbridge on the new chipsets aswell.

Intel will only lose out if Conroe is no better than the X2s. Somehow I think Conroe will be better than X2s, otherwise I see Intel going down fast.

All I'm trying to say is that a new chipset is not a big problem for Intel - that won't be their undoing in regards to point 3.
Everything else I agree with.

Sorry, I guess I wasn't clear...I was saying that it was an avoidable error by management that will cost them money and slightly increase the cost of systems. It's not a major problem, and it's easily fixed...but it shouldn't have been made and it will cost. The reason it costs Intel is that it will increase the cost of ownership by a few percentage points...not much, but in a market where margins are already razor thin it was just another mistake that they shouldn't have made (this is not AMD vs Intel, this is judging the new management's performance). Why it costs money is:

1. mobo makers must redesign...this is always costly. They must also test these new designs, print new manuals, retool their manufacturing lines, etc...
2. OEMs must do the same, as well as update all of their marketing models.
3. Since manufacturers were previously told that the mobos were already "Conroe Ready", many will probably have to redo their marketing art...

In the scheme of things, these are all small items...but they add up and they were avoidable. If it adds an extra $10 to a system price, then it was a $10 million mistake for every million systems sold...and it shouldn't have happened.

 

[Marmion]

I guess thats true, but very minor. However, I still think that because of the southbridge change, its fair to have a new chipset.
In all honesty, when AMD goes K10, mobo makers are going to have to do the same thing as what they have to do with Conroe.
Its a new architecture, so its fair enough that there is a new chipset.

Obviously, I don't agree with Intel and releasing so many chipsets for the P4 architecture.

Lets just leave it at that

 

[openwheelformula1]

Just a side note to the chipset topic from several posts above:

Current 945 and 955 chipsets will NOT support Conroe. Specific 975 and all 965 will support Conroe. Intel will also release 946 family to support single core Conroe and other 800fsb Intel CPUs.

 

[FelixDeCat]

As mentioned in my previous thread regarding this issue, I couldnt agree more.

Intel is finally leaping ahead. Those who disagree are just too backwards in their thinking.

Intel is the future, again.

 

[Viditor]

It's times like this that I wish sunlight really DID turn trolls to stone...

 

[carlosd]

Originally posted by: dmens


What's your point? Everyone knows 8xx was rushed (backpats all around).

Last time I checked, the success of a product was defined by how much money it makes.

Well it depends how well you do the marketing on the product. You can take a piece of crap and make an excellent marketing campaign , and many idiots are going to buy it. So good marketing strategies makes a succesfull product, it doesn't mean it's a good product, do I have to give an example of such products??

 

[carlosd]

Originally posted by: Cooler
The real reason Intel was in the lead in market share is that AMD PR really sucks at best.

The new intel scheme is not very clever neither for the average Joe.

 

[carlosd]

DP

 

[carlosd]

Originally posted by: FelixDeKat
As mentioned in my previous thread regarding this issue, I couldnt agree more.

Intel is finally leaping ahead. Those who disagree are just too backwards in their thinking.

Intel is the future, again.

Felixdecrap is the troll of all times. more that 7000 posts of improductive trolling, that's amazing!!, the only intelligent post you made was the one in which you considered to buy an AMD dual core :laugh:

 

[FelixDeCat]

Originally posted by: carlosd

Originally posted by: FelixDeKat
As mentioned in my previous thread regarding this issue, I couldnt agree more.

Intel is finally leaping ahead. Those who disagree are just too backwards in their thinking.

Intel is the future, again.

Felixdecrap is the troll of all times. more that 7000 posts of improductive trolling, that's amazing!!, the only intelligent post you made was the one in which you considered to buy an AMD dual core :laugh:

Just because you disagree with me trollito doesnt make you any less a troll. :laugh:

 

[FelixDeCat]

Originally posted by: carlosd

Originally posted by: dmens


What's your point? Everyone knows 8xx was rushed (backpats all around).

Last time I checked, the success of a product was defined by how much money it makes.

Well it depends how well you do the marketing on the product. You can take a piece of crap and make an excellent marketing campaign , and many idiots are going to buy it. So good marketing strategies makes a succesfull product, it doesn't mean it's a good product, do I have to give an example of such products??

The X2 was made for idiots like you! You are unworthy to own Intel carlito!